Understanding Machine Learning Categories: Supervised, Unsupervised, Semi‑Supervised, and Reinforcement

Continuing from the previous article, machine learning can be divided into four categories based on the learning paradigm: supervised learning, unsupervised learning, semi‑supervised learning, and reinforcement learning.

Supervised learning requires manually labeled training samples; each sample is assigned a known label before training, and the model iteratively improves accuracy and coverage. A typical algorithm is logistic regression.

Unsupervised learning works with unlabeled data, forcing the model to discover inherent structures such as association rules or clusters.

Semi‑supervised learning uses a mixture of labeled and unlabeled samples, enabling predictions for classification and regression tasks; support vector machines are a common algorithm.

Reinforcement learning, exemplified by AlphaGo, treats input data as feedback that directly influences the model’s behavior (e.g., scoring actions). It is applied to dynamic systems and robot control, with algorithms like Q‑learning and temporal‑difference learning.